| 1 |
On Friday 12 March 2004 07:12 pm, Luke-Jr wrote: |
| 2 |
> On Friday 12 March 2004 01:54 am, AJ Armstrong wrote: |
| 3 |
> > I'd have issues with it. The main problem is that if bash currently |
| 4 |
> > receives a directory where it is expecting a command, it fails |
| 5 |
> > transparently. |
| 6 |
> > |
| 7 |
> > In bash scripts, or complex pipelined or xarg'd one-liners, when I make |
| 8 |
> > an error that passes a directory instead of a command, I want it to |
| 9 |
> > fail. I do _not_ want it to interpret that as a cd, move to another |
| 10 |
> > directory, and carry on with what it was doing - say renaming or |
| 11 |
> > deleting. |
| 12 |
> |
| 13 |
> Perhaps it could be provided as an alternate binary meant for shell usage |
| 14 |
> only. Then any BASH scripts would continue to use the unmodified patch. |
| 15 |
|
| 16 |
no offense, but that's a dumb idea |
| 17 |
|
| 18 |
a 'good' solution has already been provided; namely adding a shopt to control |
| 19 |
this behavior ... |
| 20 |
alternatively, perhaps it could be added to bash-completion ? ive never looked |
| 21 |
at the package so i cant say whether this is possible ;) |
| 22 |
-mike |
| 23 |
|
| 24 |
-- |
| 25 |
gentoo-dev@g.o mailing list |
Archives